Climate and climate change

60

Ghana is part of a larger continental and global problem. Globally, only an estimated 16% of agricultural soils are free of significant constraints, such as poor drainage, poor nutrient status, difficult to work, salinity or alkalinity, or shallowness. Out of these favourable agriculture soils, 60% are in temperate areas and only 15% lie within tropical areas (Wood et al., 2000). Agricultural land in Africa is poor in soil fertility and quite a significant proportion is degraded to extents where recovery is uneconomical. Most soils therefore, require careful management and investments to maintain or raise productivity for supporting crop production (Coffie and Penning de Vries, 2005). This is the situation with the northern Savannah of Ghana, especially the UER as illustrated in my discussions. A decrease in soil quality along the continuum of land-use intensity in northern Ghana and its adverse implications for subsistence agriculture is observable (Braimoh and Vlek, 2005). Earlier on, they estimate that about 3% of the landscape in northern Ghana was abandoned cropland and that this was probably driven by a decline in soil fertility (Braimoh and Vlek, 2004:14). In this context, the EPA maintains that sustainability of food crop yields is closely linked with the careful management of soil fertility, especially soil organic matter and control of erosion (EPA, 2002).

61

conditions. They reach their maximum northern extent in August bringing along heavy rainfall. The influence of these air masses has resulted in a single rainy season with monthly totals increasing gradually from March and peaking around August or September and thereafter starts falling (DGRD, 1992; Yaro, 2004). Mean daily temperatures have an annual average of 28.6°C but may be as high as 32.8°C in April.

Annual rainfall varies between 700-1200 mm in the region with an average of 986 mm for Navrongo¹⁸ between 1961 and 1990 (Kranjac-Brisavljevic et al., 1999). Annual rainfall for the Guinea Savannah of the Volta basin is estimated at around 1200 mm per year (Kunstmann and Jung, 2005). See Figure 2.2 for map showing observed annual rainfall distribution in the Volta Basin in the rainy season (May- October).

Figure 2.2: Observed annual rainfall (mm) in the Volta Basin from 1992-2000

Source: Kunstmann and Jung (2005:2)

Unlike the southern part of the country, which has a bimodal rainfall pattern, northern Ghana generally, has a unimodal rainfall season annually. This has given rise to only a single major agriculture production season in northern Ghana since most agricultural

18District capital of the Kassena-Nankana District (KND).

62

production is rain fed. Given that farming is a major livelihood and irrigation infrastructure is limited, underemployment in the dry season is high among the local populace and this has adverse implications for livelihood vulnerability. Furthermore, rain fed agriculture is adversely impacted by climate change and variability. For instance, rainfall is described as erratic and changes occur from year to year. This makes predictability of rainfall and appropriate timing of planting difficult exposing crops to risk of failure (Laube, 2007). Given my interest on livelihood vulnerability, I will explore the phenomenon of climate variability in the Volta Basin with a focus on northern Ghana.

Patterns of climate variability are similar across the entire Guinea Savannah zone of the Volta Basin within which the Atankwidi lies. The rationale for this exploration is that climate variability gives rise to peculiar risks (vulnerability) that rural people confront in their livelihoods in the Atankwidi basin.

The impacts of global environmental change (Turner et al., 1990; Kasperson et al., 2001, Chapter 1) vary from region to region. According to the IPCC (2007), more droughts will occur in drier areas such as in West Africa. Aside the IPCC, many researchers acknowledge the incidence of regional „climate change‟ or preferably, climate variability in West Africa (e.g., Jung and Kunstmann, 2007; Van de Giesen et al., 2008; Kasei et al., 2010). In discussing climate variability, I will focus mainly on rainfall. This is because my empirical discussion on livelihood vulnerability (Chapters 3 and 5) are centered on rainfall. I will however, draw on changes relating to temperature and evapo-transpiration where appropriate because of their interconnectedness with rainfall in shaping local climate. A distinction between inter- annual variability and intra-annual variability is a suitable starting point. One of the most pronounced features of rainfall variability in the Volta Basin of West Africa is its inter-annual variability. Since the historic droughts of the 1970s, West Africa has been described predominantly as drought prone although a mixed scenario of dry, normal and wet years exists. Such patterns are correlated with larger scale global oceanographic and atmospheric circulation but also land surface properties of the Volta basin (Jung and Kunstmann, 2007). The 20^th century has shown a large variability in rainfall patterns in West Africa (Neumann et al., 2007; Oguntunde et al., 2006). The 1930‟s and 1950‟s were wet decades, while the 1970‟s and 1980‟s were

63

very dry decades. In particular, 1983-1984 were drought disaster years during which large parts of the rainforest including farmlands in Ghana got burnt. This dry period lead to widespread hunger in the country. Since 1990, mixed scenarios (above and below average) annual rainfall years comparable to that in the 1940‟s and before 1930 have been recorded (Oguntunde et al., 2006). The post 1970 witnessed a significant decline in average annual rainfall. From a mean of 1100 mm/year over the period (1901 -1969), it declined to 987 mm/year over the period (1970-2002). Once this single „outlier‟ is accounted for, no clear trend remains of rainfall pattern in the Volta basin (Oguntunde et al., 2006; see also Van de Giesen et al., 2008). While some research suggests a recovery of the rainfall in the sub-region (Nicholson 2005), a decline in rainfall amount and duration is observed by others for the Volta Basin. Rainfall deficiency has increased since the early 1970s, and moderate to severe drought years have occurred with a return period of approximately 9 years. High impacting droughts with areal extents of 50 % or more in the basin occurred in 1961, 1970, 1983, 1992 and 2001 (Kasei et al., 2010:89). Although rainfall is characterized by the occurrence of a mix of dry and wet years, an overall [tendency] towards a decrease in rainfall has been observed since the discontinuities of the 1960s and 1970s (Oguntunde et al., 2006). This decrease in rainfall ranged from 15 to 30 % (Kasei et al., 2010) with the 1980s being the driest period of the 20^th century in West Africa (Nicholson and Palao, 1993). At this point, I will draw extensively on drought (rainfall) analysis for the Volta basin by Kasei et al. (2010) using Standardized Precipitation Indices (SPI)¹⁹. This is because their analyses extensively focus on meteorological data on northern Ghana (including Navrongo, Atankwidi basin) (Figure 2.4). Regional climate simulation for the Volta basin by Kunstmann and Jung (2005) is also drawn on to support my discussions rainfall variability. The SPIs (Figure.2.3) show the mix trends of dry and wet years that characterize the rainfall pattern in Ghana (Volta Basin) per my earlier discussions. In Figure 2.3 (a), the SPI for Navrongo (including Atankwidi) in the KND is presented together with SPIs for Kete-Krachi, southern Ghana.

19The Standardized Precipitation Index (SPI) is a probability index that was developed to give a better representation of abnormal wetness and dryness in patterns of rainfall (Guttman, 2007). For example, severely-extremely dry year for the Volta basin (including analysis in Figure.6) will have an SPI< -2.5 and severely-extremely wet year an SPI > 2.5 (Kasei et al., 2007).

64

Figure 2.3: (a) SPI for the north (Navrongo) – south (Kete Krachi) and (b) SPI for the north –west (Bole) – east (Yendi) gradients.

Source: Kasei et al. (2010:93).

Normal rainfall years are represented by SPI values 0±0.5 between 1961 and 2005.

Beyond this, SPI values towards -2.0 and +2 represent extreme dryness or wetness respectively. For instance, SPI of -2.0 represent extreme dry conditions (drought) and that of +2 represent extreme wetness (heavy rainfall). The SPIs for Yendi and Bole (Figure 2.3 b), both located in northern Ghana show similar mixed trends of dry and wet years, although according to Kunstmann and Jung (2005), most significant trends are negative for the Volta basin. The analysis (Figure 2.3) clearly shows that rainfall in northern Ghana and the Volta basin at large is characterized by significant inter-annual variation.

Future projections about rainfall and climate change in the Volta basin based on different

„regional climate models‟ present similar mixed scenarios of dry and wet years ahead.

65

As my discussion show, some researchers predict a drying trend in the rainfall situation moving forward (eg., Oguntunde et al., 2006; Kasei et al., 2010). On the other hand, Kunstmann and Jung (2005) project a slight increase in total annual rainfall of 5%, but also a significant decrease (up to 70%) for April, which marks the transition from the dry season to the rainy season. They used a high resolution regional climate simulation model (IS92a ECHAM4 global climate scenario) based on “future climate” (time slice 2030 – 2039) and “recent climate” (1991-2000). Their predicted changes in precipitation show strong spatial variation (Figure 2.4).

Figure 2.4: Change in (a) annual and (b) April precipitation 2030-2039 vs 1991-2000(%).

Source: Kunstmann and Jung (2005:8)

The spatial distribution of annual rainfall change is shown in Figure 2.4 (a). A rough rainfall increase of +20% appear throughout the basin although spots of relatively less rainfall increments also dot across the basin. In April (Figure 2.4 (b)), a dramatic rainfall decrease of up to 70% over the entire Basin is predicted but this is also the time farmers typically start seeding (Kunstmann and Jung, 2005).

As already stated, the second domain of rainfall variability I intend to discuss is intra annual rainfall variability (ie. emerging patterns of change in annual distribution of rainfall). There are three issues of change associated with intra annual rainfall

66

distribution. These include shifting onset of the rainy season, „discordant‟ annual rainfall distribution giving rise to intermittent droughts and or high rainfall intensities and shortening rainfall seasons. The sowing season in the north of Ghana some few decades back started in April. However, the onset of the rainy season is shifting forward.

Atmospheric modeling results by Jung and Kunstmann (2007) suggest that this trend will continue into the near future. The onset of the rainy season is anticipated to shift roughly from April to May. In the 1990s, the sowing season already started in May, implying a shift (Kasei and Sallah, 1993; in Kasei et al., 2010:90). What makes the onset of the rainy season problematic is that it has also become difficult to predict with precision for better timing of sowing. According to Van de Giesen et al. (2008), there are regularly “false starts” of the rainy season. Spurious early rains deceive farmers into planting and thereafter expected follow up rains fail to sustain crops. Although farmers apply various risk management strategies losses are still major. In addition, farmers‟ experiences point to shifting trend in the onset of the rainy season in the year. In relative terms, farmers claim to sow 10 to 20 days later than their parents did. Laux et al. (2007) find statistical evidence to support the argument that the onset of rainy season is shifting forward. They find that the onset of the rains moved forward by 0.4 to 0.8 days/year in several components although the end of the rainy season and rainfall amounts remained largely unchanged. This also gives rise to high rainfall intensity within the rainy season with sometimes adverse implications for farming. What makes the shifting forward of the rains worst is the projection by Jung and Kunstmann (2007) that mean monthly rainfall for April will fall by 70% (Figures 2.4 and 2.5). If this projection is correct, it will likely affect the entire period within which the rains usually set in. This will have dying consequences for food crop farming.

Another feature of intra-annual rainfall variability is the „discordant‟ distribution of rainfall over the rainy season. This gives rise to occurrence of intermittent drought spells and high rainfall intensities that are often not in harmony with plant water requirements.

After the rains set in, the distributions over the rainy season greatly influence the growth and development of the crops (Kasei and Sallah, 1993; in Kasei et al., 2010:90).

67

Figure 2.5: Change in mean monthly precipitation 2030-2039 vs 1991-2000

Source: Kunstmann and Jung (2005:8)

However, within season dry spells (drought) are also common in northern Ghana and the Volta basin as a whole (Laux et al., 2007; Jung and Kunstmann, 2007, Kasei et al., 2010).

In northern Ghana (around Tamale), “a dry spell of 7 days can be expected to occur once a year in June and once in every 4 years in September at the peak of the rains” (Kasei and Sallah, 1993; in Kasei et al., 2010:90). In the context of drought spells in the Volta basin,

“a later sowing date than presently done is associated with better yields. This is because, on average, the impact of drought spells are less when sowing is done later (Sultan et al., 2005). On the other hand, there can also be high rainfall intensities and sometimes this also adversely affects plant growth and development. For instance, more rainfall is expected in the peak months of August and September in the Volta basin (Van de Giesen et al., 2008) but intensities can go to an extreme inimical to plant growth. Figure 2.6 shows relative drought frequency curve and intra-annual rainfall distribution for the Volta basin for 1961-2005. I will draw on this analysis for further illustrations.

68

Figure 2.6: Relative drought frequency curve and intra-annual rainfall distribution for the Volta basin for 1961-2005

Source: Kasei et al. (2010:94).

Figure 2.6(a) shows relative drought frequency curve for the Volta basin for 1961-2005.

Hence, drought intensity and frequency for the Volta basin (as earlier mentioned), shows an average return period of about 10 years for a moderate drought of probability 0.06 – 0.18 for the 44-year rainfall record (Kasei et al., 2010). This relates more to and or supports the discussion on inter-annual rainfall variability as discussed already. The analysis (2.6 b, c) relates to intra-annual rainfall distribution which is my focus here.

While (Figure b) shows monthly rainfall distributions of selected dry and wet years in

69

Tamale, (Figure c) shows the distribution functions of relative rainfall of selected years with 1997 as a normal rainfall year. Reference to Figure (b), the usual pattern of rainfall distribution is one in which rainfall amounts „pick up‟ in May and peaking in August and September. This is the normal year rainfall distribution pattern that supports optimal crop yields. Farmers expect this pattern of distribution as represented by the „normal‟ rainfall years of 1989, 1991, and 1997. However, when the rainfall fails to „pick up‟ well between May and July, and rainfall amounts fall below 800mm during the expected peak period (August – September); then the rainy season is impacted by drought. Such drought years as shown by Figure (b) are represented by the remaining five years (1961, 1970, 1983, 1992, and 2001). In general, the length of the rainy season remains same for both dry and normal years. Early rains start around March, „picking up‟ from May till September with late rains occurring in December. The rainfall season therefore, typically last 5-6 months during which about 80% of the rain falls (Kasei e al., 2010). In Figure (c), relativities in the distribution functions of the selected years (with 1997 as a normal rainfall year) are illustrated.

As the discussions show, climate variability, especially rain fall variability is a reality in northern Ghana and the Volta Basin of West Africa at large. Rainfall change manifest in both „inter-annual‟ and „intra-annual‟ variability. This makes rainfall unreliable and exposes farmers to multiple „stresses‟ and „risks‟ that create difficulties in food production and agro-related rural livelihoods. Climate variability exposes rural households to the risk of livelihood failures. Thus, the ensuing discussions address the impact of both land and soil degration, and climatic variability in the economic wellbeing of the populace in Ghana with a focus on the Atankwidi basin and northern Ghana.